Hitachi

JP1 Version 12 JP1/Performance Management User's Guide


10.1.2 Designing a cluster configuration

Performance Management can monitor operations in a cluster system.

Organization of this subsection

(1) Examining the configuration in HA cluster systems

Running the Performance Management programs on the logical host environment of an HA cluster system is referred to as logical host use.

(a) Cluster configuration of PFM - Manager

PFM - Manager can run on a logical host of a cluster system that has an active-standby configuration.

Even if a problem occurs on the executing node where PFM - Manager is executed, operation monitoring is continued by failing over to the standby node.

When PFM - Manager is used on a logical host, definition information and event data are stored on a shared disk and inherited when a failover occurs. If there are multiple instances of Performance Management on a single logical host, each instance uses the same shared directory.

The following figure shows the configuration when using PFM - Manager on a logical host.

Figure 10‒3: Cluster configuration of PFM - Manager

[Figure]

Only a single instance of PFM - Manager can be executed at a time on a single node.

(b) Cluster configuration of PFM - Web Console

PFM - Web Console can run on a logical host of a cluster system that has an active-standby configuration.

Even if a problem occurs on the executing node where PFM - Web Console is executed, operation monitoring is continued by failing over to the standby node.

If PFM - Web Console is used on a logical host, bookmark definitions are stored on a shared disks so that such information can be inherited when a failover occurs. If there are multiple instances of Performance Management on a single logical host, each instance uses the same shared directory.

The following figure shows the configuration when using PFM - Web Console on a logical host.

Figure 10‒4: Cluster configuration of PFM - Web Console

[Figure]

Only a single instance of PFM - Web Console can be executed at a time on a single node.

(c) Cluster configuration of PFM - Base

PFM - Base is compatible with cluster systems that have an active-active configuration. PFM - Base can be used on a logical host if PFM - Agent or PFM - RM is installed on the same logical host.

(d) Cluster configuration of PFM - Agent

Different cluster systems support different PFM - Agents. When the program to be monitored is running in a logical host environment, only some types of PFM - Agents will be able to run in that logical host environment.

If PFM - Agent is used on a logical host, definition information and performance data are stored on shared disks so that they can be inherited when a failover occurs. If there are multiple instances of Performance Management on a single logical host, each instance uses the same shared directory.

The following figure shows an example of a configuration that monitors an Oracle cluster system with PFM - Agent for Oracle on a logical host.

Figure 10‒5: Cluster configuration for PFM - Agent (PFM - Agent for Oracle)

[Figure]

Supplemental information:

PFM - Agent is used in a configuration that is compatible with the application that is being monitored. Therefore, there are some PFM - Agents that are used on a logical host and others that are used on a physical host. For example, PFM - Agent for Oracle is used on a logical host since PFM - Agent for Oracle monitors Oracle in a cluster configuration, on the other hand, PFM - Agent for Platform is used on a physical host to monitor the OSs on each node since PFM - Agent for Platform monitors OS performances. For details, see the chapters that describe operations on cluster systems in the appropriate PFM - Agent manual.

(e) Cluster configuration of PFM - RM

Different cluster systems support different PFM - RM products. Some PFM - RM products can also be used on a logical host. For details on whether the PFM - RM product you are using can be used in a cluster system, see the applicable PFM - RM manual.

If PFM - RM is used on a logical host, definition information and performance data are stored on the shared disks so that they can be inherited when a failover occurs. If there are multiple instances of Performance Management on a single logical host, each instance uses the same shared directory.

The following figure shows an example of the configuration when using PFM - RM for Platform on a logical host.

Figure 10‒6: Cluster configuration of PFM - RM (PFM - RM for Platform)

[Figure]

(2) Examining the configuration in load-balancing cluster systems

(a) Cluster configuration of PFM - Manager

The PFM - Manager processing cannot be balanced across multiple nodes.

PFM - Manager must be used on a physical host or HA cluster system, rather than a load-balancing cluster system.

(b) Cluster configuration of PFM - Web Console

The PFM - Web Console processing cannot be balanced across multiple nodes.

PFM - Web Console must be used on a physical host or HA cluster system, rather than a load-balancing cluster system.

(c) Cluster configuration of PFM - Base

Conforms to the configuration of PFM - Agent or PFM - RM on the same host.

(d) Cluster configuration of PFM - Agent

This subsection describes how PFM - Agent for Platform is used on each node of a load-balancing cluster system.

PFM - Agent for Platform monitors OS performances. Therefore, even in a cluster system, PFM - Agent for Platform is executed on physical hosts to monitor the OSs on each physical host. PFM - Agent for Platform must be used in the same manner as a system that is not a cluster system. Even when used in a cluster system, PFM - Agent for Platform is not registered in the cluster software.

Figure 10‒7: Cluster configuration of PFM - Agent (PFM - Agent for Platform)

[Figure]

For details on the cluster configuration of PFM - Agents, see the chapters that describe operations on cluster systems in the appropriate PFM - Agent manual.

(e) Cluster configuration of PFM - RM

Depending on the PFM - RM product you are using, it might not be possible to use PFM - RM in a load-balancing cluster system. For details, see the chapters that describe operations on cluster systems in the appropriate PFM - RM manual.